00:00welcome to the andreessen horowitz
00:02podcast I'm Hendricks Evans I'm here
00:04talking today with Steven Sinofsky about
00:07the topic that's lasted the entire
00:09history of the computing industry which
00:11is bundling and unbundling I wrote a
00:13blog post on Sunday that started out by
00:15quoting Jim Barksdale he was supposedly
00:19doing the Netscape IPO presentations
00:21told people that there are only two ways
00:22to make money in software bundling and
00:24unbundling but of course it goes back a
00:26lot longer than that and the thing that
00:28I was sort of poking away at is that
00:30when you look at what's going on in
00:32smartphones today on the one hand you
00:35have this trend towards unbundling
00:37features from inside apps and making
00:39standalone apps I think Fred Wilson
00:41called this a constellation of apps and
00:43so you have Facebook pulling stuff out
00:44of Facebook you have Foursquare
00:46splitting its app into you have lots of
00:47other people trying to do single purpose
00:49apps and the motivation for that is a
00:51limited screen real estate which changes
00:54the discovery dynamic and be that any
00:57app is just two taps away so it's much
00:59easier to switch between different apps
01:00and so you see what's happened Instagram
01:02unbundling Facebook and you see Facebook
01:05unbundling itself but you also
01:07fundamentally I think see apps
01:08unbundling the web because everything
01:10used to be just in that one icon and now
01:12to pull that into lots of icons and you
01:14know Steve you were saying this is kind
01:16of a story that goes back to your office
01:17and it goes back to mainframes and
01:18everything else the whole the whole
01:20thing just goes I mean you know it's
01:22it's in a sense it's it's just it's
01:24about innovation I mean you know you you
01:25you you have an idea to to complete some
01:29task it could be a consumer task a
01:31business task and then you build it and
01:32and then what you find is that your idea
01:35is sort of you know related to other
01:37ideas and so you start to add features
01:40and then one day you have like someone
01:44else's whole vertical integrated in to
01:46your product but it's never as good as
01:48the separate one but it might be good
01:50enough to keep your customers from
01:52jumping and using another product yeah
01:55and so you you and then one day
01:57everybody's telling you your product is
01:59too big and bloated and complicated so
02:01then you think ok well it's time to
02:03split us up into different modules and
02:07also you you know you're competitively
02:08you were frozen in time and we'll say
02:10somebody else might pull your feet right
02:12into another model in fact that's what
02:14often happens like the feature you did
02:16someone says well they're not going far
02:17enough because of course to you it was
02:19sort of a checkbox okay
02:21the best example of this is is
02:23historically it was always adding email
02:26yes like and so people would add just a
02:28little bit of email and now it's
02:29messenger for a while in the middle it
02:31was always photos and so you can look at
02:33basically every app that included photos
02:36and they all do you know a little bit of
02:39auto brightness and contrast a little
02:41bit of redeye and how of course
02:42everybody does a few filters but they're
02:44never as much as what you could get in
02:46in a full-fledged photo lab but so then
02:50those are going to keep existing and
02:52then you're either gonna completely
02:53bloat your your messenger ish kind of
02:56application with more photo editing or
02:58you're gonna lose to the photo editing
03:00people and you're sort of in this this
03:02quagmire of discoverability if you will
03:05yeah so I sort of think that like that
03:07struck me was and I was I was looking
03:08particularly at what I do has been doing
03:11with maps in China which is we've got
03:13this down out we've got this narrative
03:15that says is is endless sort of fissile
03:17Paris nests of applications that things
03:19aimlessly get kind of split apart and
03:20split apart and split apart like kind of
03:22trotskyists X the point is that all
03:29these things kind of split apart and
03:30spread apart and spread apart and then
03:31every now and then this is kind of it's
03:33almost like a kind of gravitational
03:34thing that things get kind of sucked
03:36back in again yeah like you know the
03:37cycle of the universe and but the thing
03:40that you see in China is that you have
03:42these apps at bars lots of stuff inside
03:44so just as we spit feels like this is
03:46totally inevitable totally inherent in
03:48the platform and this is just where it's
03:50going to be and then you look at China
03:51and it completely undermines all of
03:53those assumptions because what you have
03:54it is very strong portal models and
03:55particularly on mobile you have for
03:57example Baidu building a Maps app in
03:59which you can book restaurants and you
04:01can book cinema tickets and you cannot
04:03just put cinematic is but change the
04:04seat in the cinema and you can not just
04:07book a restaurant but you know order
04:08home delivery form within the app and
04:10all of those things kind of work it
04:12seems to me because they flow out of the
04:15context that is to say you know if you
04:20one might be trying to do everything
04:22that would be the web and you need
04:24vehicle in order to find things without
04:25Google you just kind of get lost in a
04:27sea of features but if you have an app
04:29that says okay it's about a location
04:31context we don't need to have 45 buttons
04:34one for restaurant someone for cinemas
04:36one for hotels one for taxis we can just
04:39like wait for somebody to do a search
04:40for restaurants right and then when they
04:42do a search for restaurant all the
04:43restaurant icons appear and when you tap
04:45on the restaurant icon what what would
04:46make sense to appear there a booking
04:48option rave and so it feels like there's
04:51this sort of the scope somewhere in the
04:54middle for integrations that focus on
04:56particular sets of use cases whether
04:58it's location which is obviously you
05:00know kind of sort of fairly logical one
05:02or alternatively social but particularly
05:06when we kind of you know the difference
05:08I think between one of the differences
05:12was through between doing this in the
05:13sort of the packaged software world and
05:14doing it in the internet world is that
05:16the big challenge for everyone is how
05:18you get content you know if you're a web
05:21publisher you're not really sitting and
05:23thinking about HTML versus apps or
05:26should it be in Facebook or should it be
05:27in this kind of Apple should it be in
05:29that kind of out what you really care
05:30about is how you get traffic and what
05:32different things you need to do in order
05:34to get that and the interesting kind of
05:37pendulum that swings back and forth is
05:39okay do I get qua traffic by having my
05:42own lap and having a deep link from that
05:43from within the Facebook app well from
05:45having my own website and getting a link
05:47to that from in the Facebook and do I
05:49expect the traffic to that website to
05:50come from Google search on a smart phone
05:52or do I have an app if I have now how do
05:54I get people to install that out are
05:56they going to find it inside the app
05:57store if so what would their search for
05:59or maybe I can get people to install the
06:01app by now within Facebook so this is
06:03just this kind of enormous match of what
06:07your options are and how where your
06:09traffic comes from well it was basically
06:11you did a website and then you were
06:13Google when you're kind of done right
06:14what's fascinating but this is that it
06:16actually opens up a really a really
06:18strong fluent for product managers to
06:20think about in terms of design because
06:22you're really dealing with a very clear
06:25measure which is usage of your service
06:28yeah so if you stick to that it's a
06:31really good thing like in fact like I
06:33could I would argue that like all the
06:35places that are downloading the number
06:36you're measuring the number of downloads
06:37on their app and things like that or
06:39that's like a secondary measure the
06:41primary measure is engagement with the
06:43service and in fact you could go too far
06:45and focus on just getting the downloads
06:47and then you'll be just victims of you
06:49know download shrapnel where like you
06:51just got a lot of downloads and no one
06:52is using your product for a whole
06:54variety of reasons and but what you have
06:57to keep in mind is that that you can
06:59continue to play out that metric as a
07:01designer and then you could start to
07:03sort of force awkward behavior like you
07:06you can drive people to try to do things
07:08in the app that are sort of for
07:10engagement as opposed to solving their
07:12problem and I think in particular
07:13because of the way that the apps have
07:17evolved in China they're they're sort of
07:19not at the monetization phase just yet
07:21they're really at just trying to drive
07:23engagement and they're solving the
07:25problems that people have and I think we
07:27have a history of websites that start
07:29off by engagement and then they try to
07:31over monetize that engagement and then
07:33the design choices sort of lose out to
07:35the the business case and it's sort of a
07:37thing to really watch and where that
07:39will manifest itself in apps is if you
07:42know you just get prompted too often to
07:44download a companion app or you know you
07:46and after like the feature breakout just
07:49seems kind of artificial like it's not
07:51really driven by screen real estate or
07:53capabilities or something it seems like
07:55you're just being launched into another
07:57place because they really wanted to
07:59break up the tap into more atomic units
08:02like if your favorite social network all
08:04of a sudden breaks out photos as a
08:06separate app because they want to be
08:08more like Instagram which seems broken
08:10out but the problem is is that they're
08:12all these things that they're gonna end
08:14and then you're just you're not even
08:15gonna be sure where to start and I think
08:18for me the whole thing that's
08:19fascinating about China is they know
08:21yeah they're starting with the place
08:23that it has my credit card and my
08:25profile and I just flow from there and
08:28yeah you're kind of you you're
08:30kind of turning the you're turning the
08:33aggregation upside down in a sense
08:35because on you know in in the US every
08:39separate app has a map built in and if
08:42it's a location thing or it has
08:44messaging built in if it's a sort of a
08:45social thing so you know Yelp has maps
08:47in it and TripAdvisor has maps in it and
08:51you know eater has maps in it and all
08:53these different things will have maps in
08:54it and uber has maps in it and lets have
08:57maps in it where was in China the Maps
09:00app has taxis and restaurants right
09:02right and the messaging app has taxis
09:05and restaurants right and indeed if you
09:07go into Baidu and you want a taxi it
09:09will show you all the cabs in your city
09:11form the five different taxi companies
09:14that are in that city and then it will
09:16bid for all of them and you whoever
09:18whoever gives the best response gets the
09:20thing first and then you also of course
09:21have a tapped call button so you can
09:24send a voice message to the driver to
09:25tell him where he is from inside the
09:26taxi module within the Maps app so this
09:28is likely you know the old um and
09:30there's supposedly this very old story
09:33in in in philosophy or astrology
09:35astronomy that um a a sort of a Hindu
09:41guru or Eastern guru is saying that the
09:43world is flat and it sits on the back of
09:45a giant turtle and somebody says well
09:47what does a turtle sit on and he says
09:49another turtle looking at some of these
09:54things it's like okay how many different
09:56layers of aggregation have we got
09:58written in this thing okay so I've got
09:59Siri or I've got people now I've got
10:01Windows Live Tiles and then I've got the
10:03Earth's layer of the the all the
10:04different I get a grid of icons and then
10:06I've got widgets and then I've got maps
10:09and I've got the social application and
10:10then within the Maps application I've
10:12got these different aggregators of
10:13restaurants or taxis that are all
10:15integrated into that and if I okay I've
10:17got the how many different ways do I
10:20you get you get all deeply philosophical
10:22but there's a more practical there not a
10:25more correct but there's a practical way
10:26that this is really going to play out
10:28which is I always like to fall back on
10:30the classic psychology paper the classic
10:33the magical number seven plus or minus
10:35two yeah flows outside the US there's a
10:37lot of history and how that is tied to
10:39why phone numbers the United States or
10:41seven digits for example and and of
10:44course now you have to remember phone
10:45numbers at all so sort of yeah for
10:47websites yeah but the idea is is that
10:50you know for better or worse like most
10:53everybody is going to just use about
10:55seven apps on their phone yep and and if
10:58the same holds for PCs I mean I remember
11:01you know we we used we did what the
11:03windows 7 launch we we talked about that
11:05there were over 800,000 apps that got
11:07used just during the beta process of
11:09course no one person used 800,000 and
11:12there were several million people in the
11:14beta and it turns out that most people
11:16use you know a single-digit number of
11:18apps if you go to all this complicated
11:20stuff you know like it's Facebook one
11:22app or two apps or three or four apps
11:24right now well it really is the the
11:28Facebook app is the one that most people
11:29will launch a very version of where
11:32people might launch the Instagram app as
11:34a standalone one but for the most part
11:36it's the thing that the seven plus or
11:38minus two is the the action on the home
11:41screen now that drives that and and
11:43that's a good thing because it means
11:45that you know an app apps like like the
11:48maps in China or commerce in China you
11:51think about launching that one app and
11:53as so long as they're good scenarios and
11:55user flow it works fine and it doesn't
11:57really matter how many other apps you
11:59use anymore so many things that
12:01integrated into it as long as they play
12:02naturally out of the pipe exactly and
12:04and the problem is really if you you
12:07start to litter the home screen or or
12:10think that the person is just going to
12:12magically remember ten different things
12:13for you it's really not gonna
12:17play out - to your advantage at all so
12:20this was I mean I think you know there's
12:22a story that in The Wall Street Journal
12:23I think over some stove a survey that
12:25most people have got the user now
12:26feature 25 apps ever and again for me so
12:28no eight at a time and I wonder how much
12:31that that aligns to how many websites
12:32people use so it's probably almost
12:33exactly the same and then and then what
12:35you have on the back end of that see we
12:37had this data from flowy and comScore
12:38earlier this year that you know as it
12:41might be 80% of people time spent on a
12:43smartphone is within an app there are
12:45maining 20% is have been in the web and
12:49so say that 80% is within those two
12:52dozen apps then you go to the web and
12:56you say okay probably 80 percent of your
12:57time was within one to two thousand
12:59websites and the rest of the web and so
13:01what happens is that Google picks up
13:03that about 20 percent which is the other
13:11980 apps or websites or whatever it is
13:14say Google picks up that vast long tail
13:16or and it probably in fact on the on the
13:19desktop internet Google also picks up
13:21the 20 things that you use every day
13:22because that's how you get to them you
13:24just go into Google any search for
13:25Facebook or you go into Google of course
13:26you're going to equal and you search for
13:27people as well right um
13:29so you've got this kind of question of
13:32okay how many different services are you
13:33using them are you using how often does
13:36that there's a mix within that changed
13:38how do you become one of those or how
13:39does one of those fall out and what is
13:42the other as it might be 500 things that
13:45you might use in the course of a month
13:47or two months and how do you get to
13:48those I mean quotes that I used in my
13:52blog post this is this famous line from
13:54Lenin that the only question the matters
13:57in politics is who whom who has the
13:59power and over whom do they exert it
14:01right and you could say the only thing
14:02that really matters in the internet for
14:04any kind of publisher or developer is
14:06say who has traffic who has a
14:08distribution who has the users and who
14:10do they give it to really and on the
14:12desktop Internet Google had them all and
14:14Facebook had them all and on the mobile
14:17Internet it's just kind of a lot more
14:19complex and kind of unsettled and
14:20confused it's a little
14:22like that I mean I think that the you
14:24know if you just take the the assertion
14:28that like it's going to be a
14:29single-digit number of things then it's
14:31just it's a matter of counting and
14:33deciding how to divide up the pie chart
14:35in terms of I don't think of it in terms
14:37of time more than anything because look
14:38you're gonna use you know probably the
14:41messaging app you're gonna use the mail
14:43app you've you know a few of us might
14:44still use the phone app like for making
14:47voice calls yeah ma'am then there's a
14:49browser which you're going to end up
14:50going to from any of the other apps that
14:52you use now here's an interesting one if
14:54you only use apps where the browser is
14:56embedded how does the OS or measurement
14:59systems even count that as an app or not
15:01an app like if you you know you're in
15:03Twitter as an app yeah and then you look
15:05click on a link and you don't click on
15:07open and Safari you click on the link
15:11and it just open it up in place you you
15:13know do a little bit of a weird penalty
15:15box but you know you have to deal with
15:17well that's as much the desk it's not so
15:18much a desk as a web it's a death of the
15:20URL right right and so to me like you
15:23could get all tangled up about like what
15:25it is you're measuring but just keep in
15:26mind that for the most part people are
15:28gonna tap on something in single-digit
15:31ways and then it's really important to
15:34remember that the tail is is essentially
15:36infinitely long like mathematically
15:38speaking yeah that there are very few
15:41apps in in either of the major app
15:43stores that have never been used and
15:46even the ones that have used by a very
15:48small number of people there are still
15:50people who use them a lot like I was
15:52talking to somebody who you know built
15:54an app for calculating some specific
15:57kind of time and photography there's
15:59probably a couple of people that use
16:01that for their workload workflow and
16:03that's what makes the app store so
16:04valuable and why you shouldn't sort of
16:06lose faith if you're not the app used by
16:09you know 30 hours a week by somebody
16:11because even if it's seven plus or minus
16:13two there's a there's a hundred percent
16:15certainty that you could be in a
16:16conference room with with ten people and
16:18each of you will have one app that the
16:20other nine people don't use so this is
16:23this reminds me I mean the thing that I
16:24thought I was going to say unkind
16:29perhaps just a realistic thing to say
16:30which is that when developers say I made
16:32this amazing app and
16:33can find it on the App Store now it's a
16:35bit like saying well I wrote this book
16:37and nobody in it's not on the home page
16:38of Amazon right I've got a million
16:40schools on the home page in the app
16:41stores and Amazon sells whatever it is
16:43two or three million books and you know
16:46you reach a point where the Internet is
16:48everything anybody on earth can make an
16:50app anybody owners can make a webpage
16:51anybody on earth can write a book and so
16:54that discovery problem is just sort of
16:57inherent in the system and that fact
16:59that most people are only going to use 5
17:01to 7 10 15 20 he's also just inherent in
17:04human nature I mean this is the point
17:07that our colleague Scott Cooper wrote
17:09this blog post about how venture capital
17:11is changing so on the one hand the cost
17:12of creating anything has collapsed by
17:14probably two orders of magnitude in the
17:15last 15 years but on the other hand the
17:17cost of you going out and getting five
17:20or six billion consumers yeah yeah
17:22haven't totally collapsed in the same
17:24way you know yeah a marginal cost has
17:26collapsed but if you want to go out and
17:28get a billion people to use your product
17:29you only need to get spectacularly lucky
17:32or you need to spend a lot of money we
17:34just realize that you can build a
17:36business out of having a half a million
17:38people use you a lot yes and and that's
17:40part of why like this you know I got to
17:42be on the the top of the you know the
17:45productivity category in the App Store
17:46or something is sort of a bit of a crazy
17:49dream now the flip side of that is it's
17:51also the same problem if you are already
17:53established and you have a bunch of
17:54customers there's this gravity that
17:56pulls all of you to R&D to be included
17:58in the app that you already have because
18:00you don't want to have to go out and
18:02attract distract your marketing and
18:04sales efforts from the big thing that
18:06you're trying to get everybody to use
18:07which is why I like in a sense and a way
18:09to wrap up right for me at least you
18:11know is the bundling unbundling you know
18:13is it a pendulum well it's not really a
18:15pendulum they're just two approaches and
18:17really they're both good unless one of
18:21them isn't good yeah well that's the
18:22thing that's kind of a trade-off between
18:24application discovery and feature
18:26discovery that if you've got a simple
18:27Apple if you've got an application with
18:29all of your stuff in it it's probably
18:31easier for you to get people to install
18:32that out but then they're never going to
18:34find all of your features what exactly
18:35is if you break stuff out into multiple
18:37apps then the feature set is really easy
18:39to discover but you've got to get people
18:41it persuade people to install those apps
18:43which is why the sort of you know it's
18:45the Facebook approach versus abide to
18:46approach you know if you stuff
18:48everything into the maths application
18:49and you do it right then you don't have
18:52a problem getting people to install body
18:53mass spec then you've got a real
18:55challenge around how we make get people
18:57to know what the hell if they can do it
18:58in the app exactly as you pull people
19:00out then you don't have a form of
19:03understanding what to do in the app but
19:04you've got a problem getting people to
19:05install a thing in the first place and
19:06when we post this one just forever
19:08I have a post that I did that talked
19:10about how to do this from a product
19:12management perspective and give some
19:13pros and cons of the different
19:14approaches for discoverability and
19:16usability so thanks a lot right thank